Pattern-recognizing self-tuning controller
Abstract
A pattern-recognizing, self-tuning controller is provided for controlling a process wherein measured characteristics including at least one peak of an error signal, derived from the differences occurring over time between the values of a process controlled variable and a desired set-point level for that variable, are used for identifying the behavior pattern of the error signal so that an operating parameter of the controller can be changed as required to minimize process recovery time whenever the process is subsequently disturbed or an abrupt change is made to the set-point level at some later time. The preferred embodiment of the device is in the form of a proportional-integral-derivative (PID) controller in which the PID coefficients are calculated in accordance with prescribed relationships that are based on damping, overshoot and time period characteristics of the error signal. Provisions are also made for including a user-specified noise threshold in order to reduce substantially the possibility of detecting a noise peak as a true peak of the error signal. A pre-adapt mode is also included, in which the controller automatically determines the initial values of the PID coefficients, the noise threshold, and the approximate time scale of the process, before on-line adaptive control of the process is given to the controller.
Claims
exact text as granted — not AI-modifiedWhat is claimed is: .[.1. A pattern-recognizing self-tuning controller apparatus for use in a control system, wherein said controller apparatus includes a controller which is coupled to a process in a closed loop of said control system for controlling a process controlled variable having set point and measured values, said controller having a control action which is determined by at least one operating parameter so that the controller is responsive to a difference between said set point and measured values; said controller apparatus comprising:
desired values is less than a prescribed difference..]. 2. .[.The apparatus of claim 1 wherein said pattern characteristic includes.]. .Iadd.A pattern-recognizing self-tuning controller apparatus for use in a control system, wherein said controller apparatus includes a controller which is coupled to a process in a closed loop of said control system for controlling a process controlled variable having set point and measured values, said controller having a control action which is determined by at least one operating parameter so that the controller is responsive to a difference between said set point and measured values; said controller apparatus comprising: first comparator means responsive to differences between said set point and measured values for generating a closed-loop behavior signal; detector means responsive to the closed-loop behavior signal of said process control variable for measuring a pattern characteristic thereof and producing a measured value of said pattern characteristic, the latter including .Iaddend.a first characteristic being a peak amplitude of a local extremum of said behavior signal and a second characteristic being an amplitude of said behavior signal at a point subsequent to said local extremum.Iadd.; second comparator means being coupled to said detector means and operating to compare said pattern characteristic measured value with a predetermined desired value for the pattern characteristic; and adjusting means coupled to said second comparator means for responding to a difference between said measured and desired values of the pattern characteristic so that the measured value substantially matches the set point value, said adjusting means not operating to alter said operating parameter of the controller when the difference between said measured and desired values is less than a prescribed difference; and .[.said controller apparatus further includes:.]. means coupled to said detector means for combining said first and second characteristics to produce a measured value of said performance criterion.
. The apparatus of claim 2 wherein said performance criterion is an overshoot ratio which is proportional to said first and second
charateristics. 4. The apparatus of claim 2 wherein said detector means measures a third characteristic of said closed-loop behavior signal of the process controlled variable, said pattern characteristic includes first and second performance criterion which have corresponding first and second desired values, said combining means generates measured values of the first and second performance criterion, and said difference means further obtains the difference between the desired and measured values of said first performance criterion and of said second performance criterion respectively, and wherein said controller apparatus further includes logic means coupled to said adjusting means for selecting between said first performance criterion difference and said second performance criterion difference so that said adjusting means responds to said selection for
altering said operating parameter. 5. The apparatus of claim 4 wherein said first performance criterion is an overshoot ratio which is
proportional to said first and second characteristics. 6. The apparatus of claim 5 wherein said second performance criterion is a damping ratio which
is proportional to said first, second and third characteristics. 7. The apparatus of claim 6 wherein said operating parameter includes a proportional (P) coefficient and an integral (I) coefficient, wherein said controller apparatus further comprises means for measuring time intervals between said first, second and third characteristics, and means responsive to said time intervals for determining a period (T 0 ) which represents the time period of the closed-loop behavior; and wherein said adjusting means changes said proportional (P) coefficient based on the selection made by said logic means and changes said integral (I) coefficient based on an adapting ratio and said measured period (T 0 ) wherein said
adapting ratio has a prescribed value. 8. The apparatus of claim 7 wherein the prescribed value of said adapting ratio is variable in accordance with predetermined relationships based on the values of said period (T 0 ), of said measured damping and overshoot ratios, and of the (I) integral coefficient existing in said apparatus which produced the closed-loop
behavior being evaluated. 9. The apparatus of claim 7 wherein said adapting ratio includes a second prescribed value and said operating parameter further includes a derivative (D) coefficient which has a value that is changed based on the second prescribed value of the adapting ratio
and on said period (T 0 ). 10. The apparatus of claim 9 wherein the first and second prescribed values of the adapting ratios are variable in accordance with predetermined relationships based on the values of said period (T 0 ), said measured damping and overshoot ratios, and the I coefficient existing in said apparatus which produced the closed-loop
behavior being evaluated. 11. The apparatus of claim 10 wherein said first and second prescribed values of the adapting ratios are Ziegler-Nichols
ratios. 12. The apparatus of claim 10 wherein changes in the proportional (P) coefficient are also related to prescribed relationships based on the measured and desired values of said damping ratio and information from a
prior cycle of said closed-loop behavior. 13. The apparatus of claim 12 wherein changes in said proportional (P) coefficient are also based on predetermined relationships which include first and second adapting ratios, said period (T 0 ), and the values of the integral and derivative (I and D) coefficients existing in said apparatus which
produced the closed-loop behavior being evaluated. 14. The apparatus of claim 13 wherein said first, second and third characteristics are only measured if the absolute value of said first characteristic is greater
than a predetermined noise level. 15. The apparatus of claim 14 further including means for measuring the time occurrence of said first characteristic and means for changing said time occurrence by an interval based on a subsequent time when the absolute value of the amplitude of
said closed-loop behavior signal falls to a prescribed level. 16. The apparatus of claim 15 wherein said prescribed level is equal to 95 percent
of the measured amplitude of said first characteristic. 17. The apparatus of claim 15 wherein said measured period (T 0 ) is further adjusted based on relationships which include the measured amplitudes of said first, second and third characteristics, said time intervals between said first, second and third characteristics, a fourth characteristic based on the interval between said time occurrence of the first characteristic and a time when the absolute value of said closed-loop behavior signal falls to a second prescribed level, and to the time intervals between the occurrences of previous first, second and third characteristics of a prior
cycle of said closed-loop behavior signal. 18. The apparatus of claim 17 wherein said second prescribed level is equal to 60 percent of the
measured amplitude of said first characteristic. 19. The apparatus of claim 17 wherein said first, second and third characteristics are three local extrema, and the absolute values of the amplitudes of the extrema are greater than associated levels that are each based on a predetermined
noise level. 20. The apparatus of claim 17 wherein said first, second and third characteristics are three local extrema, and the absolute values of the amplitudes of only the first two extrema are greater than associated
levels that are each based on a predetermined noise level. 21. The apparatus of claim 17 wherein said first, second and third characteristics are three local extrema, and said fourth characteristic is less than the fourth characteristic of a previously evaluated cycle of said closed-loop
behavior signal. 22. The apparatus of claim 17 wherein only said first and second characteristics are respectively two local extrema of said closed-loop behavior signal and the absolute values of the amplitudes of the extrema are greater than associated levels that are each based on a
predetermined noise level. 23. The apparatus of claim 17 wherein only said first characteristic is a local extrema of said closed-loop behavior signal and the fourth characteristic is less than the fourth characteristic of a previously evaluated cycle of said closed-loop
behavior signal. 24. The apparatus of claim 17 wherein said second and third characteristics both occur at a time equal to a prescribed maximum time interval from the time occurrence of said first characteristic, and said fourth characteristic is greater than zero and less than the fourth characteristic of a previously evaluated cycle of said closed-loop
behavior signal. 25. The apparatus of claim 17 wherein said first and second characteristics are local extrema of said closed-loop behavior signal, and said third characteristic occurs after a prescribed maximum interval from the time occurrence of said first characteristic, and said
fourth characteristic is greater than zero. 26. The apparatus of claim 17 wherein a search for said second characteristic begins after a first prescribed search interval from the time occurrence of said first characteristic and a search for said third characteristic begins after a second prescribed search interval for the time occurrence of said first
characteristic. 27. The apparatus of claim 26 wherein said first prescribed search interval is based on the period (T 0 ) divided by 4 and said second prescribed search interval is based on the period (T 0 ) divided by 4 and the time occurrence of said second
characteristic. 28. The apparatus of claim 17 wherein a search for said second characteristic begins after the occurrence of said first characteristic and after a time when the absolute value of the amplitude of said closed-loop behavior signal has fallen to a prescribed search
level. 29. The apparatus of claim 28 wherein the prescribed search level
is equal to 50 percent of the amplitude of said first characteristic. 30. The apparatus of claim 17 wherein said detector means further includes means for identifying short duration peaks in said first, second and third characteristics as noise peaks, and means coupled to said noise identifying means for removing said noise peaks prior to the time
occurrence of each of said first, second and third characteristics. 31. The apparatus of claim 30 further including means responsive to an open-loop behavior signal of said process controlled variable for
determining an initial value for said operating parameter. 32. The apparatus of claim 31 further including means responsive to a steady state condition of said process controlled variable for measuring said noise
level. 33. In a pattern-recognizing self-tuning controller device which is coupled to a process in an open loop mode for controlling a process-controlled variable having a succession of values at associated instances of time, said controller device having at least one operating parameter which affects how it responds to a change of an input thereto, wherein said process is responsive to a control signal produced by said controller device, that improvement in said controller device which comprises: means for generating a trace signal which represents a time plot of said succession of values of the process-controlled variable, slope detecting means being responsive to the trace signal resulting when said control signal is appropriately bumped at a time T from one level to another level so that said process controlled variable correspondingly changes from one steady state condition having a first time base line to a new steady state condition having a second time base line, said slope detecting means including: means for determining the slopes of lines extending from a point T f which is fixed on said first time base line to successive value points of said trace signal as said process controlled variable changes from said one to said new steady state conditions; means for selecting from said lines a line having a maximum absolute value of slope; means for identifying a value point of said trace signal through which said maximum slope line passes as an upper inflection point; means for selecting an increment point of said trace signal, said increment point occurring prior to said upper inflection point; means for measuring absolute value of the slope of a process indicative line which extends through said upper inflection point and said increment point; means for measuring the intercept of said process indicative line with said first time base line; means for measuring the time difference between said intercept and said time T and producing a first signal representative of a process dead time characteristic T dt of said process; and means responsive to said slope of the process indicative line for producing a second signal representative of a process sensitivity characteristic SEN; and adjusting means coupled to receive the output of said slope detecting means for responding to said first and second signals to alter said controller
device operating parameter. 34. The apparatus of claim 33 wherein said increment point means further selects a second point of said trace signal, said process indicative line is chosen from first and second lines which extend respectively through said upper inflection point and said first and second points; and said detector means further includes: means for selecting between said first and second lines the line having the maximum absolute value of slope, said maximum slope line being set as the
process indicative line. 35. The device of claim 34 wherein said operating parameter includes proportional (P), integral (I) and derivative (D) coefficients and said adjusting means operates in accordance with predetermined relationships which include the following: Proportional (P) coefficient equals 120 times T dt divided by SEN; Integral (I) coefficient equals 1.5 times T dt Derivative (D) coefficient equals (I) Integral coefficient divided by 6; and
Period (T 0 ) equals 5 times T dt . 36. The apparatus of claim 35 further including: means for determining a noise level characteristic of said trace signal; and means coupled to said noise means for altering said derivative (D) coefficient based on a predetermined relationship which is proportional to
said noise level characteristic. 37. The apparatus of claim 36 wherein said noise means comprises: high-pass filtering means for removing from said trace signal prescribed high-frequency components; integrator means coupled to receive the output of said filtering means for integrating said output over a prescribed time interval; averaging means responsive to the output of said integrtor means for generating a noise band signal having a value 2NB which is representative of the peak-to-peak noise band characteristic of said trace signal and which is based on a prescribed average value of said integrator means
output. 38. The apparatus of claim 37 wherein said predetermined relationships are modified based on a quantity Z where Z=(3.0-2NB)/2.5 and based on the following if 0<Z<1, derivative (D) coefficient equals Z multiplied by the quotient of integral (I) coefficient divided by six, and
if Z<0, derivative (D) coefficient=0. .Iadd.39. A method for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising the steps of: generating an error signal equal to the difference between the process variable signal and a setpoint signal; measuring a first amplitude of a first local extremum of said error signal; measuring a second amplitude of a second local extremum of said error signal; combining said first measured amplitude and said second measured amplitude to produce a measured performance characteristic; and automatically adjusting at least one of the control parameters of the controller to minimize the difference between said measured performance characteristic and a selected value of said performance characteristic. .Iaddend. .Iadd.40. The method of claim 39 wherein said step of measuring a second amplitude includes measuring a second local extremum which is opposite in direction, adjacent, and later in time to said first local extremum. .Iaddend. .Iadd.41. The method of claim 40 wherein said step of combining includes producing an overshoot value as said measured
performance characteristic. .Iaddend. .Iadd.42. The method of claim 40 further comprising the step of measuring a third amplitude of a third local extremum of said error signal. .Iaddend. .Iadd.43. The method of claim 42 wherein said step of measuring a third amplitude includes measuring a third local extremum which is opposite in direction, adjacent, and later in time to said second local extremum. .Iaddend. .Iadd.44. The method of claim 43 wherein said step of combining includes combining said third measured amplitude with said first and second measured amplitudes to produce said measured performance characteristic. .Iaddend. .Iadd.45. The method of claim 44 wherein said step of combining includes producing a damping value as said measured performance characteristic. .Iaddend. .Iadd.46. The method of claim 44 wherein said step of combining includes producing two measured performance characteristics. .Iaddend. .Iadd.47. The method of claim 46 wherein said step of combining includes producing a damping value as one of said measured performance characteristics and producing an overshoot value as the other of said measured performance
characteristics. .Iaddend. .Iadd.48. A method for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising the steps of: generating an error signal equal to the difference between the process variable signal and a setpoint signal; measuring a first amplitude of a first local extremum of said error signal; measuring a second amplitude of a second local extremum of said error signal; combining said first measured amplitude and said second measured amplitude to produce a measured performance characteristic; comparing said measured performance characteristic to a selected value of said performance characteristic to generate a performance characteristic difference; and automatically adjusting at least one of the control parameters of the controller to minimize said performance characteristic difference. .Iaddend. .Iadd.49. The method of claim 48 wherein said step of measuring a second amplitude includes measuring a second local extremum which is opposite in direction, adjacent, and later in time to said first local extremum. .Iaddend. .Iadd.50. The method of claim 49 wherein said step of combining includes producing an overshoot value as said measured
performance characteristic. .Iaddend. .Iadd.51. The method of claim 49 further comprising the step of measuring a third amplitude of a third local extremum of said error signal. .Iaddend. .Iadd.52. The method of claim 49 wherein said step of measuring a third amplitude includes measuring a third local extremum which is opposite in direction, adjacent, and later in time to said second local extremum. .Iaddend. .Iadd.53. The method of claim 52 wherein said step of combining includes combining said third measured amplitude with said first and second measured amplitudes to produce said measured performance characteristic. .Iaddend. .Iadd.54. The method of claim 53 wherein said step of combining includes producing a damping value as said measured performance characteristic. .Iaddend. .Iadd.55. The method of claim 53 wherein said step of combining includes producing two measured performance characteristics. .Iaddend. .Iadd.56. The method of claim 55 wherein said step of combining includes producing a damping value as one of said measured performance characteristics and producing an overshoot value as the other of said measured performance
characteristics. .Iaddend. .Iadd.57. A method for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising the steps of: generating an error signal equal to the difference between the process variable signal and a setpoint signal; measuring amplitude values of at least two local extrema of said error signal; utilizing said amplitude values to produce a measured damping characteristic of said error signal; and automatically adjusting at least one of the control parameters of the controller to minimize the difference between said measured damping characteristic and a selected value of said damping characteristic. .Iaddend. .Iadd.58. The method of claim 57 wherein one of said control parameters is a proportional coefficient and wherein said step of automatically adjusting includes adjusting said proportional coefficient. .Iaddend. .Iadd.59. A method for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising the steps of: generating an error signal equal to the difference between the process variable signal and a setpoint signal; measuring amplitude values of at least two local extrema of said error signal; utilizing said amplitude values to produce a measured damping characteristic of said error signal; comparing said measured damping characteristic to a selected value of said damping characteristic to generate a damping characteristic difference; and automatically adjusting one of the control parameters of the controller to
minimize said damping characteristic difference. .Iaddend. .Iadd.60. The method of claim 59 wherein one of said control parameters is a proportional coefficient and wherein said step of automatically adjusting includes adjusting said proportional coefficient. .Iaddend. .Iadd.61. A method for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising the steps of: generating an error signal equal to the difference between the process variable signal and a setpoint signal; measuring amplitude values of at least two local extrema of said error signal; measuring the times of occurrence of said amplitude values; computing a period from said times of occurrence; utilizing said amplitude values to produce a measured damping characteristic and a measured overshoot characteristic of said error signal; and automatically adjusting at least one of the control parameters of the control to minimize the difference between a measured ratio of said parameter to said period and a selected value of said ratio in response to said measured overshoot characteristic and said measured damping characteristic. .Iaddend. .Iadd.62. The method of claim 61 wherein said control parameters include an integral and a derivative coefficient and wherein said step of automatically adjusting includes adjusting said integral and said derivative coefficients in response to the difference between said measured damping characteristic and said measured overshoot characteristic. .Iaddend. .Iadd.63. An apparatus for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising: means for generating an error signal equal to the difference between the process variable signal and a setpoint signal; means for measuring a first amplitude of a first local extremum of said error signal; means for measuring a second amplitude of a second local extremum of said error signal; means for combining said first measured amplitude and said second measured amplitude to produce a measured performance characteristic; and means for automatically adjusting at least one of the control parameters of the controller to minimize the difference between said measured performance characteristic and a selected value of said performance
characteristic. .Iaddend. .Iadd.64. An apparatus for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising: means for generating an error signal equal to the difference between the process variable signal and a setpoint signal; means for measuring a first amplitude of a first local extremum of said error signal; means for measuring a second amplitude of a second local extremum of said error signal; means for combining said first measured amplitude and said second measured amplitude to produce a measured performance characteristic; means for comparing said measured performance characteristic to a selected value of said performance characteristic to generate a performance characteristic difference; and means for adjusting at least one of the control parameters of the controller to minimize said performance characteristic difference. .Iaddend. .Iadd.65. An apparatus for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising: means for generating an error signal equal to the difference between the process variable signal and a setpoint signal; means for measuring amplitude values of at least two local extrema of said error signal; means for utilizing said amplitude values to produce a measured damping characteristic of said error signal; and means for automatically adjusting at least one of the control parameters of the controller to minimize the difference between said measured damping characteristic and a selected value of said damping characteristic.
.Iaddend. .Iadd.66. An apparatus for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising: means for generating an error signal equal to the difference between the process variable signal and a setpoint signal; means for measuring amplitude values of at least two local extrema of said error signal; means for utilizing said amplitude values to produce a measured damping characteristic of said error signal; means for comparing said measured damping characteristic to a selected value of said damping characteristic to generate a damping characteristic difference; and means for automatically adjusting at least one of the control parameters of the controller to minimize said damping characteristic difference. .Iaddend. .Iadd.67. An apparatus for automatically adjusting the control parameters of a self-tuning controller of a process having a measured process variable signal, comprising: means for generating an error signal equal to the difference between the process variable signal and a setpoint signal; means for measuring amplitude values of at least two local extrema of said error signal; means for measuring the times of occurrence of said amplitude values; means for computing a period from said times of occurrence; means for utilizing said amplitude values to produce a measured damping characteristic and a measured overshoot characteristic of said error signal; and means for automatically adjusting at least one of the control parameters of the controller to minimize the difference between said measured period and a selected value of said measured period in response to said measured overshoot characteristic and said measured damping characteristic. .Iaddend.Join the waitlist — get patent alerts
Track USRE33267E — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.